bioRxiv preprint first posted online May. 4, 2018; doi: http://dx.doi.org/10.1101/314658. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.
1 2 3
Direct reprogramming of human epithelial cells into organoids by miR-106a-3p
4 5
Delom F.1*, Puceat M.2, and Fessart D.1*
6 7 8
1 Univ. Bordeaux, F-33000 Bordeaux, France.
9
2 Université Aix-Marseille, INSERM UMR_1251, F-13385 Marseille, France.
10 11 12
* Corresponding author:
13
Delphine Fessart - Bordeaux University - 146, Léo-Saignat – 33076 Bordeaux – France. Phone:
14
(00 33) +5 56 33 04 31. Fax: (00 33) +5 56 33 33 30. E-mail:
[email protected]
15
Frederic Delom - Bordeaux University - 146, Léo-Saignat – 33076 Bordeaux – France. Phone:
16
(00 33) +5 56330431. Fax: (00 33) +5 56333330. E-mail:
[email protected]
17 18
Lead contact: Delphine Fessart,
[email protected]
19 20 21
1
bioRxiv preprint first posted online May. 4, 2018; doi: http://dx.doi.org/10.1101/314658. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.
22
Abstract
23
Organoids development relies on the self-organizing properties of adult stem cells to create structures
24
which recapitulate the architecture, functionality, and genetic signature observed in original tissues.
25
Little is known about of the exact nature of the intrinsic cell properties at the origin of organoid
26
generation, and of the signaling pathways governing their differentiation. Herein, we carried out a
27
functional microRNA screen to identify miRNAs at the origin of organoid generation from human
28
epithelial cell culture. We uncover miR-106a-3p that initiates and promotes organoids. This miRNA
29
acts as a master inducer of the expression of the three core pluripotency transcription factors
30
(NANOG, OCT4 and SOX2) through the regulation of a set of 10 genes, and thus strengthening the
31
reprogramming and cell differentiation of human epithelial cells into organoids. These data
32
demonstrate that organoids can be directly generated from human epithelial cells by only one miRNA:
33
miR-106a-3p. Hence, we appear to have identified a new determinant of organoid identity, which
34
plays a role in reprogramming, cell differentiation and tissue engineering.
35
2
bioRxiv preprint first posted online May. 4, 2018; doi: http://dx.doi.org/10.1101/314658. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.
36
Introduction
37
Three-dimensional (3D) human organoid culture models are appealing tools to study
38
pathophysiological processes. These models have been described, by us and others, for the
39
lung
40
like, reflecting the ability of organoid culture conditions to prompt cells to self-organize into
41
structures mimicking the architecture of the organ from which they were derived. In contrast
42
to organ explants, organoids can arise from a single primary cell
5,6
as well as for numerous other tissues
19
. The term "organoid" literally means organ-
5,6,28
, thereby allowing the
5
43
generation of human organoids from biopsies . Non-tumor organoids are thought to arise
44
from adult stem cells (aSCs), and therefore should in theory be capable of self-renewal and
45
differentiation. Consistent with this idea, primary cells enriched with known progenitor/stem
46
cell markers are more efficient at forming organoids than the general cell population
47
However, currently there is a lack of understanding of the underlying epigenetic and genetic
48
mechanisms that control organoid-initiating frequency, self-renewal and differentiation during
49
organogenesis.
13
.
50
To better understand mammalian development, as well as to exploit the tremendous
51
therapeutic potential of organoid models, it is necessary to identify and characterize the
52
genetic mechanisms governing the fate of aSCs. MicroRNAs (miRNA) have been recently
53
shown to play an important role in regulating stem cell self-renewal and differentiation 15. In
54
general, one gene can be repressed by multiple miRNAs and one miRNA may repress
55
multiple target genes, which results in the formation of complex regulatory networks. In a
56
wide variety of developmental processes, miRNAs finely tune or restrict cellular identities by
57
targeting important transcription factors or key pathways
58
the contribution of miRNA-mediated gene regulation in the enrichment of progenitor/stem
59
cell markers. This would allow us to better characterize the mechanisms responsible for
60
controlling the initiating cell sub-population and thus improving tissue-specific organoid
61
growth conditions. Consequently, we performed a miRNA screen in human primary epithelial
62
cells to identify the mediators influencing the initiation of stem cell derived organoids. We
63
identified a previously uncharacterized miRNA, miR-106a-3p, and its target genes that play a
64
key role in such process. Using a gain of function approach, we discovered that the
65
endogenous levels of three core transcription factors (OCT4, SOX2 and NANOG) were post-
66
transcriptionally controlled by miR-106a-3p in human aSCs. Moreover, we discovered that
67
miR-106a-3p is necessary and sufficient to fine tune the differentiation process, and thus the
68
pluripotent state through a specific transcriptional regulatory network. Overall, our results
20
. Hence, we sought to investigate
3
bioRxiv preprint first posted online May. 4, 2018; doi: http://dx.doi.org/10.1101/314658. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.
69
highlight the importance of miR-106a-3p in the initiation of stem-cell derived organoids and
70
provide some clues about the mechanism underlying organogenesis.
71
4
bioRxiv preprint first posted online May. 4, 2018; doi: http://dx.doi.org/10.1101/314658. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.
72
Results
73
Organoid culture of Human Epithelial Cells exhibits a CD44high/CD24low phenotype
74
This study was initiated to identify organoid-initiating epithelial cell subpopulations 13
75
that specify stem/progenitor cell functions in epithelial cells
76
characteristics of stem cell is to be rare immortal cells within a mass culture that can both self-
77
renew by dividing and give rise to many cell types. First, we characterized the properties of
78
human primary mammary epithelial cells (HMEC) grown in 3D compared to conventional 2D
79
culture. As a control, cells were grown under organoid culture conditions as we previously
80
described
81
Approximately 3% of cells present in the culture featured the capacity to reconstitute an
82
organoid (Figure 1A), suggesting the presence of stem cells within the mass culture. Next, the
83
self-renewal capacity of organoid-initiating cells was assessed by serial organoid formation
84
from passage 5 to passage 11 (Figure 1B). Cells progressively lost self-renewal ability to form
85
organoids upon serial propagation (Figure 1B), consistent with previously described loss of
86
self-renewal potential of primary epithelial stem cells after few passages 14.
5
. One of the main
and the cell lines tested formed 3D-structured human organoids (Figure 1A).
87
Previous studies have reported that human mammary epithelial cells with
88
CD44high/CD24low phenotype have the highest progenitor ability compared to all other
89
stem/progenitor subpopulations 7. Therefore, we analyzed the expression of CD44 and CD24
90
in 2D cell culture compared to 3D using flow cytometry (Figure 1C). In 2D cell culture, 85%
91
expressed both CD24 and CD44 at high levels and 14% expressed CD24 at low levels
92
together with high levels of CD44 (Figure 1C, Top panel). In contrast, 3D cell culture showed
93
more than 3-fold increase in CD44high/CD24low phenotype cells (~49%) compared to 2D
94
(~14%) (Figure 1C; lower panel, p=0.0268, n=3). We then analyzed the cells from 2D culture
95
using standard immunofluorescence (Figure 1D) to determine the expression of CD24 and
96
CD44 cell-surface markers. The overlaid images showed a mix of cell populations: CD24
97
cells (green), CD44 cells (red) and CD44/CD24 co-expressing cells (yellow) (Figure 1D).
98
Together, these results indicate that cells grown as organoids acquired a CD44high/CD24low
99
expression pattern similar to stem/progenitor cell that can be used for further screening.
100 101
A miRNA screening approach to selectively favor organoids formation
5
bioRxiv preprint first posted online May. 4, 2018; doi: http://dx.doi.org/10.1101/314658. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.
102
To investigate whether miRNA-mediated gene regulation could promote organoid
103
formation, we monitored, as a tool, the expression of CD44 and CD24 following miRNA
104
transfection into HMEC cells (Figure 2A-D). Following quantitative image analysis of
105
>100,000 cells at Passage 6 (P6), frequency distributions of CD44 intensity were compared in
106
mass culture (whole population), and mass culture exposed to CD44 siRNA (Figure 2A) or
107
exposed to CD24 siRNA (Figure 2B). CD44 and CD24 levels were lower in siRNA-depleted
108
cells in comparison to the whole population, which validates the specificity of our assay
109
(Figure 2A-B). To identify miRNAs that play a role in the enrichment of CD44high/CD24low
110
cell phenotype, we performed an unbiased functional screen for miRNAs that modulate
111
CD44/CD24 phenotypes in HMEC (Figure 2C). Using an approach similar to our genome-
112
wide small interfering RNA (siRNA) screen for p16 modulators 2, we transfected actively
113
proliferating cells (Passage 6, P6) with 837 miRNAs. siRNA targeting siGLO (‘cyclophilin
114
B’; PPIB), CD44 or CD24 served as controls. We assigned cut-off values to define miRNA
115
hits based on CD44 and CD24 cell density. The raw screening data and quantitation of each
116
phenotypic criterion are shown in Figure 2D. This strategy revealed that the miR-106a-3p
117
shifts primary cells into a CD44high/CD24low phenotype. This miRNA is a paralogue of the
118
miR-17/92 cluster (Figure 2E). Next, to further confirm the results, we performed a secondary
119
screen of the whole family cluster (Figure 2E). Twenty-eight miRNAs belonging to the
120
cluster were retested, in triplicate, using the same method as in the primary screen (Figure
121
1C). A total of 4 hits were scored as those miRNAs with Z-factor >2 (Figure 2F) and
122
prompted a shift in the CD44high/CD24low population (Figure S1B). The top hit was miR-
123
106a-3p (Figure 2F and Figure S1A). We then confirmed miR-106a-3p induced a
124
CD44high/CD24low phenotype using flow cytometry based on the expression of CD44 and
125
CD24 (Figure S1B and Figure 2G). In cells expressing the control mimic, the
126
CD44high/CD24low population (CD24-/CD44+) was ~10% of the total cell population
127
(Figure 2G). Conversely, we observed a 5-fold increase of CD44high/CD24low population,
128
~50% of the total cell population, in cells transfected with a mimic miR106a-3p (Figure 2G).
129
In parallel, to correlate these data with organoid development, we assessed the
130
organoid-initiating frequency of each of the 28 miRNAs (miR-17/92 cluster) (Figure 2E). Out
131
of the total of 7 positive hits (Figure 2H), miR-106a-3p displayed the highest organoid-
132
initiating frequency (Figure 2H). Taken together, these results show that miR-106a-3p (Figure
133
2I), is the only miRNA that exhibits the two properties of 1) enriching CD44high/CD24low
134
cells and 2) favoring organoid development. 6
bioRxiv preprint first posted online May. 4, 2018; doi: http://dx.doi.org/10.1101/314658. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.
135
The development of human organoids is driven by miR-106a-3p.
136
Next, we questioned whether miR-106a-3p is endogenously expressed in organoids
137
compared to 2D culture. We found that only the 3D culture of organoids expressed miR-106a-
138
3p, thus reinforcing its potential role in organoid formation (Figure 3A). To further study
139
miR-106a-3p function, we generated retroviral vectors of miR-106a as previously described 24
140
evaluated its stable expression in HMECs (Figure 3B-E). First, we examined the expression of
141
miR-106a-5p and miR-106a-3p using RT-qPCR in control (miR-Vector) and miR-106a-
142
infected cells (Figure 3B) and observed that miR-106a-5p was expressed in both conditions.
143
On the contrary, both RT-qPCR and in situ hybridization showed that miR-106a-3p was
144
exclusively expressed in miR-106a cells (Figure 3B-C) at levels similar to those observed in
145
3D cultures (Figure 3A).
146
As expected, miR-106a stable overexpression greatly increased organoid-initiating
147
frequency (Figure 3D). To further evaluate the impact of miR-106a on organoid architecture,
148
organoids were analyzed using confocal microscopy. Apoptotic cells are present in organoids
149
during lumen development
150
Caspase-3, demonstrated that miR-106a did not impact on luminal apoptosis during
151
organogenesis (Figure 3E, Caspase-3). Moreover, organoids are characterized by a well-
152
defined cell/Matrigel interface with a myoepithelial layer, which was not impacted by miR-
153
106a overexpression (Figure 3E, CD44 and p63). In addition, organoids expressed β-catenin
154
(cell junction marker) adjacent to the plasma membranes and miR-106a overexpression did
155
not show any effect on its localization and did not disrupt cell junctions (Figure 3E, β-
156
catenin). These results demonstrate that miR-106a does not disrupt the structure of organoids.
4
. Immunofluorescence staining for the apoptosis marker,
157
To test miR-106a-3p and miR-106a-5p individual functions on capacity of organoid-
158
initiating cells, miR-106a-3p or miR-106a-5p mimics were transfected in HMEC cells (Figure
159
3F-G). We examined miR-106a-5p and miR-106a-3p expression by RT-qPCR and observed
160
that miR-106a-5p is expressed both in control, miR-106a-5p and miR-106a-3p cells (Figure
161
3F). As expected, miR-106a-3p is only expressed in miR-106a-3p transfected cells (Figure
162
3F). Next, we studied the individual role of miR-106a-5p and miR-106a-3p overexpression on
163
organoid-initiating frequency (Figure 3G). The overexpression of miR-106a-3p significantly
164
increases organoids number by about 5-fold compared to control and miR-106a-5p (Figure
165
3G). Our results indicate that miR-106a-5p does not impact on organoids frequency,
166
demonstrating the specific requirement of miR-106a-3p to mediate the self-renewal capacity 7
bioRxiv preprint first posted online May. 4, 2018; doi: http://dx.doi.org/10.1101/314658. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.
167
of organoid-initiating cells (Figure 3G). Taken together, these results demonstrate that i) miR-
168
106a-3p expression's is only restricted to organoids, and ii) miR-106a-3p, but not miR-106a-
169
5p, is required for the development of human organoids.
170
Identification of miR-106a-3p targets
171
Since a single miRNA can potentially target hundreds of genes
10
, we next cross-
172
referenced the predicted targets of miR-106a-3p using four different algorithms (miRanda,
173
miRDB, DianaMT and miRWalk). We found 67 genes common to the four algorithms (Figure
174
4A). In parallel, to study the effect of the over-expression of miR-106a-3p on global gene
175
expression patterns, we isolated total RNA from HMEC cells transfected with miR-106a-3p
176
mimic and performed microarray analysis (HG-U133 Plus 2.0). The results indicated that
177
transfection of miR-106a-3p induced significant changes in the expression of 6465 genes (p
178
value